Evolutionary algorithms for single and multicriteria design optimization (Q5950265)

A solution to a design optimization problem can be regarded as the satisfaction of a list of constraints, while optimizing several, potentially conflicting, objectives, often expressed as criteria. This book discusses various approaches of solving this difficult problem by applying evolutionary/genetic algorithms. Starting with a population of candidates expressed as binary strings, a genetic algorithm selects some of those candidates with an evaluating function, then generates new candidates by applying genetic operators, either crossover, where it swaps parts of the bits between a pair of strings; or mutation, where it randomly changes certain bits of a string. This process will be repeated until a preset condition is met. The ``best string'' produced at the end of this process is declared as the result. The evolutionary algorithm works in basically the same way, except that it allows more flexible coding mechanisms for the candidate representation, more sophisticated selection algorithms to select candidates for expansion, and more genetic operators, which are essentially variants of the two aforementioned operators, to generate new candidates. The introductory Chapter 1 discusses some of the basic concepts, such as constraints, objective functions, and the formulation of the design optimization problems, (p)reviews some of the mathematical concepts, such as various optimum notions, particularly, the Pareto optimum, and the expected ones on convexity and concavity, and reviews some of the representative conventional methods for both single- and multi-criteria optimizations. Chapters 2 and 3 cover the basic concepts of genetic, and evolutionary, algorithms. They both start with a general discussion of the concepts, which are followed by demonstrating examples, and then their applications to solve rather involved problems. Chapters 4 and 5, the main part of this book, discuss, respectively, how to apply evolutionary algorithms to solve single- and multi-criteria optimization problems. In both cases, it overviews the relevant issues, presents a survey of the known methods to address these issues, and supplies examples to demonstrate some of those mentioned techniques. Chapter 6 covers other evolutionary algorithm based methods. For example, a method of selecting a representative subset of Pareto optimal solutions in the case of multi-criteria optimization is discussed. Chapter 7 ends the book by presenting several engineering examples to demonstrate the effectiveness of the EOS algorithm contained in the first appendix, which implements some of the methods as discussed in the book and is coded in the programming language \(c\). The second appendix describes two design optimization problems, also in \(c\). This book contains an impressive list of references, 283 in total, a list of symbols used within this book, as well as an quite helpful index. This book is nicely written and well organized. Besides a clear discussion of all the concepts, the examples are well chosen, easy to follow, and real. I particularly like the fact that those examples are hierarchically structured in the sense that an easy one is followed by a more advanced one, which is in turn followed by a few inspiring applications found in the real life. This structure certainly serves very well the stated intention of this book, namely, a self-study guide for researchers, and students. I highly recommend this book to any one who not only wants to learn and understand this fascinating technique of evolutionary/genetic algorithms, but also wants to find out its application in the real life, particularly, in the very fitting area of solving engineering design optimization problems.